Le Chat has gained Deep Research, a mode that plans queries, searches the web, and assembles reference-backed briefs in seconds. Released on 17 July 2025 and live for all web and mobile users without payment, it targets analysts, students, and product teams who need rapid due diligence. The same update introduces speech input powered by Voxtral Small (24 B) and Voxtral Mini (3 B); users can dictate ideas or record meetings with low delay.
BREAKING 🚨: Mistral AI is releasing a bunch of new features! Deep Research, Voice Input, Projects, Image Editing and Native Multimodal Reasoning.
— TestingCatalog News 🗞 (@testingcatalog) July 17, 2025
Feature overload! 🔥 https://t.co/sdKY0OAhi4 pic.twitter.com/aZGxEyA1EG
Magistral now drives “Think” responses, delivering longer, multilingual reasoning that can switch languages mid-sentence. New Projects folders keep chats, files, and chosen tools together so long-running work streams retain context, while image generation gains prompt-based editing that can delete items or move subjects without resetting the scene.
Early comments on social channels praise Deep Research speed and Voxtral’s natural cadence, and enterprise pilots note that Projects narrows the gap with dedicated knowledge-management suites.
Mistral AI, founded in Paris in 2023, has raised over €1 billion to build open-weight and proprietary models such as Mixtral, Codestral, and now, the Magistral and Voxtral families. Backed by a Microsoft Azure partnership, the company positions Le Chat as a European, privacy-conscious alternative to US chatbots; today’s feature set pushes the assistant further toward that goal.
